Chat Memory

v1.0.0

Your agent forgets everything after compaction? This fixes it. Built by the AI Advantage community — the world's leading AI learning platform (aiadvantage.ai...

0· 93·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
Purpose & Capability
The skill name and description claim to convert session transcripts into searchable Markdown, build a glossary, and suggest cron optimizations. The included scripts perform exactly those tasks (convert JSONL sessions to Markdown, build a SESSION-GLOSSAR.md from those transcripts, and analyze cron job prompts). Required resources (local session files and a workspace) align with the stated purpose.
Instruction Scope
The SKILL.md instructs running the provided scripts and creating cron jobs to keep memory up-to-date. The scripts read session logs from the user's ~/.openclaw directories, write Markdown under a workspace memory folder, and produce reports — all consistent with the declared feature set. Note: SKILL.md suggests scanning ~/.openclaw/agents/*/sessions/ and documents an --agent flag, but session-to-memory.py's implementation hardcodes ~/.openclaw/agents/main/sessions and only recognizes --new/--force (no --agent argument). This documentation vs code mismatch may cause unexpected behavior unless you adjust the script or paths.
Install Mechanism
No install spec or external downloads are present; this is an instruction-only skill with local Python scripts. Nothing is fetched from external URLs or written into system-wide locations beyond the user workspace/home directories.
Credentials
The skill requests no secrets or environment variables. build-glossary.py optionally respects a WORKSPACE env var (defaults to ~/.openclaw/workspace). The scripts read user session files (which may contain sensitive content like credentials, messages, or personal data) and write index and markdown files under the user's workspace. Access to these files is necessary for the skill's purpose, but you should be aware that all session contents will be processed and persisted locally.
Persistence & Privilege
The skill does not request always:true, does not modify other skills' configurations, and has no installation step that forces permanent platform-level presence. It simply provides scripts the user runs (or runs via cron if the user chooses).
Assessment
This skill appears to do what it says: convert local OpenClaw session logs to Markdown, build a glossary, and suggest cron prompt improvements. Before installing/running: - Review the scripts locally (you already have the sources) and run them in a safe/test environment first. They process all session files found under your home (~/.openclaw) and will persist converted transcripts and indexes to your workspace. - Be aware of sensitive content: session logs often contain tokens, credentials, or private messages. The scripts will read and write that content to memory/*.md files. If you have secrets in sessions, consider removing or redacting them before indexing, or run the tool on a sanitized subset. - Note the SKILL.md vs implementation mismatches: SKILL.md documents scanning all agents and an --agent flag, but session-to-memory.py hardcodes ~/.openclaw/agents/main/sessions and only supports --new/--force. If you have multiple agents, update the script or move files accordingly. - If you plan to set up cron jobs, test manually first. The cron-optimizer only suggests prompt modifications and does not modify jobs automatically, but adding memory preambles to automated prompts can surface private context to automated tasks — review each suggested change before enabling. - Limit file permissions on the generated memory/ files if other users/processes share your machine; deleting the generated memory files will remove the local index. Given the code is local, has no network calls, and matches its stated function, it is coherent — just pay attention to privacy and the small docs-to-code inconsistencies described above.

Like a lobster shell, security has layers — review code before you run it.

latestvk97ewcsa2s1gvj6vb1kjdggdch83nqqr

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

Comments